Integrated FCC Riser - Regenerator Dynamics

studied in a Fluid Catalytic Cracking Pilot Plant

G. M. Bollas* and I. A. Vasalos

Laboratory of Petrochemical Technology, Department of Chemical Engineering, Aristotle

University of Thessaloniki, PO Box 361, 57001, Thessaloniki, Greece

A. A. Lappas, D. K. Iatridis, S. S Voutetakis

Chemical Process Engineering Research Institute (CPERI), Centre for Research and

Technology Hellas (CERTH), PO Box 361, 57001, Thermi-Thessaloniki, Greece

S. A. Papadopoulou

Automation Department, Alexander Technological Educational Institute of Thessaloniki, P.O.

Box 14561, GR54101, Thessaloniki, Greece

* Corresponding author. Chemical Process Engineering Research Institute, Centre for

Research and Technology Hellas, 6th km. Charilaou – Thermi Rd., PO Box 361,

Thessaloniki, GR-570 01, Greece, Fax: +30 2310 498380, E-mail: gbollas@cperi.certh.gr

Abstract

In this paper a dynamic simulator of the FCC pilot plant, operated in Chemical Process

Engineering Research Institute (CPERI), is presented. The simulator was developed and

verified on the basis of steady-state and dynamic experiments. The operation of the pilot plant

permits the execution of case studies for recording of the dynamic responses of the unit, by

imposing substantial step changes in a number of the manipulated variables. The comparison

between the dynamic behavior of the unit and this predicted by the simulator, arise useful

conclusions on both the similarities of the pilot plant to commercial units along with the

ability of the simulator to depict the main dynamic characteristics of the integrated system.

The simulator predicts the wt% feed conversion, the wt% coke yield and the heat consumed

by the catalytic reactions in the FCC riser on the basis of semi-empirical models developed in

CPERI and simulates the regenerator according to the two-phase theory, with a dilute phase

model in account for post-combustion reactions. The riser and regenerator temperature, the

stripper and regenerator pressure drop and the composition of the regenerator flue gas are

measured on line and are used for verification of the ability of the simulator to predict the

dynamic transients between steady states in both open- and closed-loop unit operation. All the

available process variables such as the reaction conversion, the coke yield, the carbon on

regenerated catalyst and the catalyst circulation rate are used for the validation of the steady

state performance of the simulator. The results reveal the ability of the simulator to predict

accurately the operation of the pilot plant in both steady state and dynamic conditions. The

dynamic simulator can serve as the basis for the development of a model based control

structure for the pilot plant, besides its use as a tool for process optimization studies.

Keywords: Fluid Catalytic Cracking, Mathematical Modeling, Dynamic Simulation, Pilot

Plant, Catalyst Deactivation, Bubble Columns

1

Introduction

The dynamic simulation of the Fluid Catalytic Cracking (FCC) process is a challenging

research subject of high economic and environmental importance. Optimization of this

complex process demands the development of accurate models capable of describing the

process in detail. FCC technology is continuously evolving during the last half century,

though the requirements for stable operation of commercial FCC units restrict the possibility

of obtaining accurate models over an extensive operating range, through experimentation. In

industry the target is maximum capacity (i.e. profitability), and that bounds the evidence of

dynamic transients within particular and narrow operating windows. Thus, FCC pilot plants

are often used for the simulation of commercial units under different operating conditions,

feed properties and catalyst activity and selectivity. The operation of a pilot unit provides the

ability to examine the process under steady feed and/or catalyst properties, in order to isolate

their respective effects on the cracking reactions and develop correlations for each subset of

process variables. One other asset of FCC pilot plants is the potential to examine their

dynamic behavior within a great range of operating conditions, which enhances the

investigation of the process dynamics.

The research interest in dynamic simulation of the fluid catalytic cracking process has been

consecutively increasing during the last years. Lee and Groves (Lee & Groves, 1985)

proposed a dynamic model which treats the riser as a pseudo-steady state adiabatic plug flow

reactor and the regenerator as a continuous stirred tank reactor with no dilute phase. Elnashaie

et al. (Elnashaie, Abasaeed & Elshishini, 1995; Elnashaie & Elshishini, 1993) developed a

dynamic model for an industrial type IV FCC unit and investigated the sensitivity and

stability of the system. They used two-phase models for both the reactor and the regenerator

and included unsteady state dynamic terms for the thermal behavior and for the carbon mass

balance throughout the entire unit. Lopez-Isunza (Lopez-Isunza, 1992) presented a dynamic

model for mass and energy balance, but their model neglects the hydrodynamic aspects of the

2

FCC unit. McFarlane et al. (McFarlane et al., 1993) presented a comprehensive model for the

simulation of a type IV FCC unit, in which they included the reactor, regenerator, blowers, U-

bends, compressors, furnace and valves, as to be able to compute the pressure balance and

catalyst circulation rate of this type of unit. In the regenerator model they included a dilute

phase in account for post combustion, but the riser part used oversimplified computations for

the heat balance. Arbel et al. (Arbel et al., 1995; Arbel, Rinard & Shinnar, 1995) developed a

model able to describe both the steady-state and dynamic behavior of an FCC unit. Their riser

model was based on the widely known 10-lump model (Jacob et al., 1976), assuming pseudo-

steady state conditions, while the regenerator model included a complete description of both

full and partial combustion kinetics. They extensively studied the steady state multiplicities of

the FCC unit and the effect of combustion mode on the controllability of the unit. Ali and

Rohani (Ali & Rohani, 1997; Ali, Rohani & Corriou, 1997) presented a dynamic model, in

which they developed analytical solutions of the differential model equations after adopting

pseudo-steady state assumptions. Their model neglects the freeboard region of the

regenerator. In-Su Han et al. (Han & Chung, 2001a, , 2001b) presented a detailed dynamic

simulator of the FCC process, in which they included the simulation of catalyst liftlines,

stripper, feed preheater and cyclones. They applied a distributed parameter 4-lump model for

the riser reactor and a two-regime, two-phase model for the regenerator. The UOP type fluid

catalytic cracking unit was in detail simulated by Cristea et al. (Cristea, Agachi & Marinoiu,

2003). The simulator of Cristea et al. is based on the model of McFarlane et al. (McFarlane et

al., 1993) including models for the feed preheater, the main fractionator, the air blower, the

wet gas compressor. Cristea et al. implemented model predictive control (MPC) algorithm to

their simulator and studied the effect of the control structure on the unit performance.

Recently, Hernandes-Barajas et al. (Hernandez-Barajas, Vazquez-Roman & Salazar-Sotelo,

2006) presented another dynamic simulator of the FCC unit, with a detailed pressure balance

3

of the unit. Hernandes-Barajas et al. focused on the multiplicity of steady states in fluid

catalytic cracking units.

In this paper the development and verification of a dynamic simulator, on the basis of

steady-state and dynamic experimental data of the FCC pilot plant of Chemical Process

Engineering Research Institute (CPERI), will be presented. The term “dynamic experiments”

is used to describe experiments, in which a step change is imposed to a manipulated process

variable, while recording the transient of a number of process variables from the original

steady state to the final steady state the system will eventually reach. The development of the

dynamic simulator of the pilot plant serves two main goals: a) the study of the dynamic

behavior of the pilot process that includes the validation of the model performance against

steady-state and dynamic unit responses, the identification of the process dependencies and

uncertainties, and the performance of experimental case studies to examine the similarities of

the pilot plant with commercial units, and b) the use of the simulator for the development of a

model-based optimizer and control scheme for the entire unit. This paper deals with the

former of the aforementioned goals and examines the ability of the simulator to provide an

accurate representation of the unit qualitatively and quantitatively, as well as the ability of

both the pilot plant and the simulator to simulate the main dynamic characteristics of a typical

commercial FCC process.

4

Experimental Setup

The FCC pilot plant of CPERI (Fig. 1) operates in a fully circulating mode and consists of a

riser reactor, a fluid bed regenerator, a stripper and a liftline. The riser reactor operates at

pseudo-isothermal plug flow conditions and consists of a large-diameter bottom region

(mixing zone) (26mm i.d., 0.3m height) and a smaller-diameter (7mm i.d., 1.465m height) top

region connected by a conical-shaped region 0.05m of height. At the reactor bottom, the gas-

oil contacts the hot catalyst (which flows from the regenerator) and evaporates, while the

catalyst is kept in a fluidized state by means of nitrogen flow. The cracking product from the

riser top enters the stripper vessel for the separation (stripping) of gas from catalyst. The

stripped mixture flows through a heat exchanger for condensation of the heavier compounds.

Thereupon, the mixture is led to a stabilizer column for better separation of the liquid and

gaseous products. The mixture of gasoline, light cycle oil and heavy cycle oil is obtained

through the bottom of the stabilizer. The yield to liquid products is measured with the ASTM

D-2887 simulated distillation method. The stripped catalyst flows through the liftline to the

regenerator, where the majority of the carbon, deposited on the catalyst surface, is burned off.

The regenerator consists of two main sections. A small-diameter bottom section (77.92mm

i.d., 0.715m height) and a larger-diameter top section (254.6mm i.d., 0.64m height) connected

by a conical-shaped section 0.205m of height. A standpipe at the bottom of the regenerator

leads the regenerated catalyst back to the riser bottom to continue the operation loop. Two

slide valves, one at the exit of the standpipe and one at the exit of the stripper regulate the

catalyst circulation throughout the unit. The standpipe slide valve controls the catalyst

circulation for constant riser temperature, while the stripper slide valve operates for constant

stripper pressure drop (i.e. stripping volume). Two wet test meters and two gas

chromatographers measure the volumetric flow rates and the molar composition of the flue

and cracked gas, respectively. An on-line oxygen analyzer monitors the excess of oxygen to

ensure thorough catalyst regeneration. The process control system of the unit is based on a

5

special industrial computer control system. The system is coordinated with the FIX/DMACS

S/W by Intellution. The control system collects the values of the inputs and drives the output

signals, as well as maintains a digital record of the signals. The process pressure control

valves and the power to electrical heaters are controlled by numerous PID controllers.

6

Model Description

The simulator of the pilot plant includes three main sections: a pseudo-steady state model of

the riser reactor, a dynamic model of the regenerator and a set of dynamic and pseudo-steady

state models of the stripper, the standpipe, the liftline and the slide valves. For the specific

case of the pilot plant, the dynamic effects of the riser, the cyclones, the liftline and the

standpipe on the performance of the integrated unit were neglected, because their operation

has significantly lower impact on the process dynamics, compared to the two large vessels of

the pilot plant, the stripper and the regenerator. The behavior of the regenerator dominates on

both the dynamic and the steady state behavior of the integrated unit. This is due to the

adiabatic nature of the system in which the need to balance coke formation and combustion is

the predominant force (Arbel et al., 1995). The riser residence times are much shorter

compared to the response times of the regenerator, hence one can at any instance describe the

riser reactor by a set of pseudo-steady state equations, which simplifies the dynamic analysis.

The main impact of the riser operation on both the dynamic and steady state behavior of the

integrated system is on the coke production and on the heat consumption. Thus, the accurate

prediction of pseudo-steady state conversion, coke yield and heat of cracking and

vaporization is the main concern, when describing the effect of riser in the integrated dynamic

system.

The pseudo-steady state and dynamic sub-models that constitute the dynamic simulator, as

developed in CPERI, have been presented in the literature (Bollas et al., 2002; Bollas et al.,

2004; Faltsi-Saravelou & Vasalos, 1991; Faltsi-Saravelou, Vasalos & Dimogiorgas, 1991)

and will be briefly adduced in this section. A kinetic-hydrodynamic model was developed for

the simulation of the pilot riser reactor in pseudo-steady state conditions (Bollas et al., 2002).

The catalyst hold-up and residence time in the reactor were calculated on the basis of

empirical hydrodynamic correlations and the gas-oil conversion and coke yield were

predicted through a Blanding type (Blanding, 1953) kinetic model. The prediction of gas-oil

7

conversion and coke yield are the only two lumps of the riser sub-model, essential for

inclusion in the complete simulator, which reduces the necessity of using a more detailed

lumped model. The effect of feedstock properties on gas-oil conversion and coke yield was

expressed through semi-empirical correlations developed on the basis of experiments

performed with constant catalyst and a variety of feedstocks (Bollas et al., 2004). The effect

of catalyst type was expressed through a “catalyst index” (Bollas et al., 2004). The model of

the regenerator is based on the two-phase theory (Davidson, Clift & Harrison, 1985), in which

the gas-solids flow is assumed to follow the bubbling bed regime, consisting of two zones: a

dense zone at the regenerator bottom comprised by a bubble and an emulsion phase, and a

dilute zone at the regenerator top, called the freeboard. The model equations were grouped

into two main modules that serve for the two main sections of the unit, the riser and the

regenerator. A third module was used for the simulation of the stripper and the slide valves,

the liftline and the standpipe.

Simulation of Riser Reactor

The pseudo-steady state model of the FCC riser reactor was developed on the basis of the

following assumptions:

the aggregate effect of operating conditions, feed properties and catalyst type on the

cracking reactions are simulated by the product of their discrete functions

the riser reactor is assumed to run in concurrent plug flow of gas and solids at pseudo-

isothermal conditions

second-order rate apparent kinetics are applied for gas-oil conversion (x)

catalytic coke (cx) deposition parallels catalyst deactivation (Voorhies, 1945).

On the basis of these assumptions and after integration and rearrangement of the

corresponding spatial equations eqs.(1) and (2) were formulated:

8

( ) ( ) :RS exp100

xnx xC

g RX

k Ex C catalyst type F feed quality tx WHSV

⎛ ⎞−= ⎜ ⎟⎜ ⎟− ⎝ ⎠R T

(1)

( ) ( ) :RS exp cnc cx c c C

g RX

k Ec C catalyst type F feed quality tWHSV R T

⎛ ⎞−= ⎜ ⎟⎜ ⎟

⎝ ⎠ (2)

The adjustable parameters (kx, kc, Ex, Ec, nx, nc) of eqs.(1) and (2) were estimated on the

basis of a dataset of steady state pilot experiments performed with constant feed and catalyst

quality, in a great range of space velocities (WHSV) and catalyst residence times (tC:RS) and at

two different reactor temperatures (TRX) (Bollas et al., 2002). A large database of experiments

with different feedstocks and catalysts was used for the development of models of the effect

of feedstock quality and the assignment of “catalyst indices” that are used in eqs.(1) and (2).

The methodology of the simulation of the effect of feed and catalyst on conversion and coke

yield is shortly described in Appendix B. The values of the parameters of eqs.(1) and (2) are

given in Table 1.

Finally, a pseudo-steady state heat balance of the riser reactor was performed. The main

contributors to the overall enthalpy balance in an FCC plant are: (a) the enthalpy of cracking

ΔHcrack; (b) the enthalpy of vaporization of the gas-oil feedstock; and (c) the enthalpy content

of various process streams (gas-oil, catalyst, cracked products, inerts). An empirical

correlation was developed to estimate the heat of cracking in the riser reactor. This correlation

was based on experiments performed at different temperatures, using various feedstocks and

at different conversion levels. The final correlation estimates the heat of cracking as a

function of conversion, riser temperature and gas-oil molecular weight, as shown in eq.(3):

( ) (21 2 3 1 2 3ln

100crack RX RX F RX RX FxH a T a T a MW bT b T

x⎛ ⎞Δ = + + + + +⎜ ⎟−⎝ ⎠

)2 b MW (3)

The enthalpy content of gas-oil vapors was estimated by integration of the empirical

correlation of Kesler and Lee (Kesler & Lee, 1976). The values of the parameters of eq.(3) are

presented in Table 1.

9

Simulation of Regenerator

The structure of the physical model of the regenerator is shown in Fig. 2. The fluidized bed

includes two zones: (a) the dense bed and (b) the freeboard. The dense bed consists of a

bubble and an emulsion phase, while the freeboard contains the entrained catalyst particles

that are recycled to the emulsion phase of the dense zone via cyclones.

The assumptions made for the simulation of each phase shown in Fig. 2 are (Faltsi-

Saravelou & Vasalos, 1991; Faltsi-Saravelou, Vasalos & Dimogiorgas, 1991):

the bubble phase is free of catalyst particles

plug flow regime is assumed for the bubble phase

the emulsion phase gas and catalyst particles are assumed fully mixed

the freeboard is modeled as an ideal plug flow reactor

the catalyst particles are hydrodynamically represented by their average size, density and

porosity, while the particle size distribution is used for the emulsion to freeboard

entrainment rate calculation

diffusion in the catalyst particles is neglected

due to the high temperatures in the FCC regenerator, the ideal gas law is valid

the fluidized bed reactor is adiabatic.

The velocity of the gas flowing through the emulsion phase, was assumed to be equal to the

minimum bubbling velocity, which is a consistent assumption for Geldart group A

particles (Geldart, 1973) (in which category FCC catalysts typically belong). The clouds and

wakes around the bubbles were assumed to have zero volume. This assumption is valid for

high ratios of superficial gas velocity over minimum fluidization velocity, which is typical for

operations of group A particles. The bubbles were assumed to grow in size with bed height,

while the variation of the fluidizing gas density and superficial velocity, due to axial

temperature gradients and gas molar rate changes, was also taken into account.

10

The dense bed of the regenerator was simulated as a pseudo-steady state PFR (bubble

phase) in parallel to a dynamic CSTR (emulsion phase). The dense bed volume was

calculated on the basis of the overall regenerator dynamics:

( ) ( ) ( ) ( )

( )1 0 1

:RG :RG :RG :CY :CY

1

D D F Fl l l lD C C C C

p e e

dV W W W Wdt fρ ε

= = =− + −=

−

0=

(4)

The material balance for gas components in the bubble phase is:

:RG

1 homoib

Mi b ij RjjD D

dF K f a KV dl

= − + ∑ b (5)

The energy balance in the bubble phase is given by eq.(6):

( ):RG

1 homob

H b RjjD D

dQ K f H KV dl

= − + −Δ∑ Rjb (6)

In the emulsion phase the material balance equations were formulated separately for gas

and solids components, as shown in eqs.(7) and (8) respectively:

( ) ( )

( )0 10

:RG 0

1D Dl l homo hete

geie ie iee e Mi D e e ij Rje e e ij Rje

j jge D

Wdc c cf K dl f a K f a Kdt V

ε ερ

= = −= + + + −∑∫ ε ∑ (7)

( )( ) ( ) ( ) ( )

( )11 1 1

:RG :CY

:RG :RG

1 1FD D F ll l l hete

if ieie C ie ie Ce e e e ij

jp D p D

c cdc W c c WRjef f a

dt V Vε ε

ρ ρ

== = = −−− = + + − ∑ K (8)

The energy balance equation in the emulsion phase is given by eq.(9):

( ) ( ) ( ) ( ) ( ) ( ) ( )

( ) ( ) ( ) ( )

1 0 1 0:RG:RG :RG :CY :CY

11

:RG :RG :RG0

1

1

D D F F D

D

gassolidsl l l l lD e

e e ie ie e ie ie C C C C gei i

homo hetel

0

ge loss D H D e e D Rj Rje e e D Rj Rjj j

d V Tf c cp f c cp Q Q Q Q Q

dt

Q Q V K dl f V H K f V H K

ε ε

ε ε

= = = = =

=

⎛ ⎞− + = − + − +⎜ ⎟

⎝ ⎠

− − + + −Δ + − −Δ

∑ ∑

∑ ∑∫ e

(9)

The superficial bubble gas velocity for the dimensionless fraction of dense bed height lD, is

evaluated by differentiating the ideal gas law in terms of the bubble enthalpy rate term:

:RG :RG

gb g b

D D D gb D

du R dQdl A P cp dl

= (10)

The bubble-emulsion mass interchange MiK and the heat interchange HK and the emulsion

fraction ef are evaluated by eqs. (11) - (13), respectively:

11

:RG

ibMi ti ie

gb D

FK K cu A

⎛ ⎞= ⎜⎜

⎝ ⎠− ⎟⎟ (11)

( )H t b eK H T T= − (12)

( )1

0

1e b Df f dl= −∫ (13)

The combined bubble to emulsion gas interchange coefficients are evaluated by eq.(14):

1 1b

ti bci cei

fK k k

= + (14)

For the evaluation of the bubble-cloud (kbci) and cloud-emulsion (kcei) gas interchange

coefficients the expressions proposed by Kunii and Levenspiel (Kunii & Levenspiel, 1977)

were adopted. The same method (Kunii & Levenspiel, 1977) was used for the estimation of

the heat interchange coefficient (Ht).

The freeboard is simulated as an ideal two-phase PFR. The material balances of the gas and

solid components in the freeboard are shown in eqs. (15) and (16), respectively:

( ):RG

1 1homo hete

iff ij Rjf f ij Rjf

j jF F

dFK

V dlε α ε α= + −∑ ∑ K (15)

( ):RG

1 1hete

iff ij Rjf

jF F

dFK

V dlε α= − ∑ (16)

The energy balance for the freeboard is:

( ) ( ) ( ):RG

1 1homo hete

ff Rj Rjf f Rj Rjf

j jF F

dQH K H K

V dlε ε= −Δ + − −Δ∑ ∑ (17)

The ideal gas law is differentiated in terms of the gas enthalpy rate to evaluate the gas

superficial velocity:

:RG :RG

gf g

F F F gf F

du R dQdl A P cp dl

= gf (18)

The derivative of the enthalpy of the gas phase is obtained by eq.(19), assuming that the

heat capacity of the components is constant at each integration step:

12

gf gf

F f

dQ Q dQdl Q dl

= f

F

(19)

The chemical species considered to be involved in the reaction scheme of the regenerator

are categorized to gas components (N2, O2, CO2, CO, H2O) and solid components (Al2O3,

SiO2, C, H, S). A short description of the kinetics of the heterogeneous and homogeneous

reactions is given in (R1)–(R6):

The intrinsic carbon combustion on the catalyst surface corresponds to a couple of

reactions producing CO and CO2 with second order kinetics:

[ ][ ]1

2 1 11C O CO C O2

K

r K+ → = 2

2

(R1)

[ ][ ]2

2 2 2 2C + O CO C OK

r K→ = (R2)

The homogeneous CO oxidation in the gas phase, at which the water acts catalytically:

[ ] [ ][ ]3 0.5 0.5

2 2 3 3 2 21CO + O CO O CO H O2

K

r K→ = (R3)

The catalytic CO oxidation, at which part of the CO produced on the catalyst site, is

catalytically oxidized on the catalyst itself, or on an oxidation promoter:

[4

2 2 4 41CO + O CO CO2

K

r K→ = ] (R4)

The hydrogen combustion on the catalyst surface, which produces a significant amount of

heat, is:

[ ][ ]5

2 2 5 512H + O H O H O2

K

r K→ = 2

2

(R5)

The coke sulfur combustion on the catalyst, which produces mainly SO2, is:

[ ][ ]6

2 2 6 6S + O SO S OK

r K→ = (R6)

The reaction of C and CO2 on the catalyst site producing CO is neglected, as it occurs at a

very low rate. The parameters of the kinetic expressions (R1)-(R6) are presented in Table 2.

13

Simulation of Stripper, Standpipe, Liftline and Slide Valves

The stripper was simulated as a perfectly mixed reactor in minimum fluidization conditions,

the reactor dense bed volume of which and the material balance for the solids components

were expressed through eqs.(20) and (21):

( ) ( )

( )1 0

:ST :ST :ST

1

D Dl lD C C

p mf

dV W Wdt ρ ε

= =−=

− (20)

( ) ( )( )( )

1 1:ST :ST :ST:ST

:ST1

D Dl lC i ii

p mf D

W c cdcdt Vρ ε

= = −=

− (21)

The stripping efficiency of the pilot stripper was assumed 100%, as the stripper volume and

the stripping steam flow are adequately large for the pilot riser capacity. In the pilot plant the

temperature of the catalyst at the stripper dense bed is regulated by electrical heaters, that

operate to achieve a set point value. The temperature of the catalyst stream at the exit of the

standpipe (riser entrance) and at the exit of the liftline (regenerator entrance) is calculated by

eq.(22):

( )w

C C

T TdT DLdl cp W

π−= (22)

Model Structure – Initial and Boundary Conditions

The dynamic material and energy balance equations form a system of integro-differential

equations that is solved following an iterative procedure commencing from the initial and

boundary conditions of the system, as shown in Fig. 3. The common case is that the simulator

is used to study the transient from a simulated steady state of the unit to a new one, when a

step change is imposed to one ore more of the manipulated variables. Otherwise, the system

variables receive the starting values of a “guess” steady state, estimated on the basis of the

assumption of plug flow conditions throughout the regenerator. Thereafter, the system is

solved until convergence to a valid steady state. For time zero the emulsion variables

14

( ( ) ( ) ( )0 0:RG, , t t t

ie D ec V T= = 0= and all other time dependent variables shown in Fig. 3) hold the values of

the initial steady state. The superficial gas velocity at the regenerator entrance, ( )0:RGDl

gu = , is

calculated using the ideal gas low for combustion air flow rate ( )0:RG

DlgW = at temperature ( )0Dl

bT =

equal to the air preheat temperature and pressure ( )0:RGDl

DP = equal to the regenerator bottom

pressure. The superficial gas velocity in the bubbles at the entrance of the regenerator is then

calculated by eq.(23), where ( )tgeu is the superficial gas velocity in emulsion at real time t (t = 0

for the initial steady state):

( ) ( ) ( )0 0:RG

D Dl l tgb gu u= == − geu (23)

For the dilute phase, the boundary conditions at the dimensionless height lF = 0 (end of the

dense zone entrance to the freeboard region) are:

( ) ( ) ( )0 1:RG

F Dl l tif ib ie ge DF F c u A= == + (24)

( ) ( ) ( )0 0:RG

F Fl t lif ie sf FF c u A= == (25)

( ) ( ) ( ) ( ) ( ) ( ) (0 1 0 1:RG :RG

F D F D

gassolidsl l t l t l )f b ie sf F ie ie ge D ie e r

i i

Q Q c u A cp c u A cp T= = = =⎛ ⎞= + + −⎜ ⎟

⎝ ⎠∑ ∑ T (26)

( ) ( ) ( )( )0 1 :RG

:RG

F Dl l t Dgf gb ge

F

Au u uA

= == + (27)

The catalyst with concentration ( )tiec enters the riser with rate that is determined by the slide

valve at the end of the regenerator standpipe (eq.(28)) after time lag given by eq.(29):

( ) ( ) ( )( )( )1

1

1

0.52 2 0.5SP :SV 1 1

:RS SV RG RG SP RS2 2SP :SV

2 1 F Ft l lC p b

t

A AW k P P P P P

A Aρ ε = =⎛ ⎞

= − + Δ + Δ −⎜ ⎟⎜ ⎟−⎝ ⎠RS−Δ (28)

( ) ( )( )

RG RS SP0

:RG

1D

p bdead l

C

Vt

Wρ ε→

=

−= (29)

The same formulation is used to calculate the catalyst mass flow rate entering the

regenerator after time lag determined by the residence time of the catalyst in the liftline:

15

( ) ( ) ( ) ( )( )( )2

2

2

0.52 2 0.5ST :SV1 1

:RG SV ST ST RG LL2 2ST :SV

2 1D F Ftl lC p mf

t

A AW k P P P P

A Aρ ε= =⎛ ⎞

= − + Δ −⎜ ⎟⎜ ⎟−⎝ ⎠

1l = − Δ (30)

( ) ( )( )

LLSP RG0

:ST

1D

p mfdead l

C

Vt

W

ρ ε→=

−= (31)

The industrial practice for profitable and constant operation of the FCC unit is to control the

riser exit temperature. The automatic control of the reactor temperature was included in the

simulator with a routine that adjusts the catalyst circulation rate for constant riser temperature.

Using the pseudo-steady state model of the riser, the catalyst circulation rate is regulated by

solving the system provided by the heat balance (eq.(3)) and the conversion and coke yield

equations (eqs.(1),(2)) simultaneously, at each solution cycle, as shown in Fig. 4. In Fig. 4 the

regenerator temperature, the riser temperature, the coke on regenerated catalyst, the rate and

quality of the feedstock, and the inerts rate are used for the calculation of the catalyst

circulation rate that redeems the mass and energy balances in the riser. Based on the new

calculated value for the catalyst circulation rate and the estimate of coke production, the

stripper entrance variables and the regenerator exit flow are updated and the loop continues,

until the convergence that declares the imposition of steady state is achieved.

16

Model Application to the pilot plant of CPERI

The responses of the simulator were recorded for step changes in the flow rate and preheat

temperature of the feed. For the needs of this particular study experiments were performed

with the controller of reactor temperature by catalyst circulation rate set to operation or not, in

order to examine the open and closed loop behavior of the unit. Accordingly, the part of the

simulator that adjusts the catalyst circulation rate for constant riser temperature was set active

or inactive. The open loop experiments were used for validation of the mass and energy

balances formulation and for verification of the structure of the integrated model (assumption

of pseudo-steady state operation of the riser, iterative procedure of convergence etc.). In the

open loop experiments the actions of the controller of riser temperature-catalyst circulation

rate do not interfere with the process dynamics and the net dynamic responses of the unit to

the changes in the manipulated variables can be observed. The closed loop experiments were

performed to examine the effect of the controller of riser temperature-catalyst circulation rate

on the behavior of the unit and were compared with the dynamic model responses, with the

routine of adjustment of catalyst circulation rate being active. All experiments were

performed with constant feedstock and catalyst, the properties of which are presented in Table

3.

In the next paragraphs the four cases examined are presented. First, the open loop responses

of the unit and the simulator to a 15% decrease in feed flow rate are presented. Second, the

responses of the simulator are compared with those of the unit for the same change, but in

closed loop operation. The last two cases examine the effect of increasing the feed preheat

temperature to 130% of its original value in open and closed loop operation. At this point it

should be noted that the pilot plant shows a great deviation from the ideal instantaneous step

change that the operator orders. The 15% decrease in feed rate was established in the real time

operation of the pilot plant within a 5 min period; while an average of 80 min was required

for the 130% increase of the feed preheat temperature. The main duty of the pilot plant of

17

CPERI is to perform steady state experiments in standard conditions for the evaluation of new

catalysts, thus the observed high delays in imposing changes in the manipulated variables is

understandable and justified. At this point the development of the dynamic simulator focuses

on the study of the integrated riser-regenerator system; hence it does not include models for

the feed mass flow meter and the heater. The transients in the imposition of the changes in

question were, however, recorded by the control system of the pilot plant. Thus, it was

possible to reproduce the transients of change in the manipulated variables in five consecutive

representative steps, as shown in Figures 5(a)-8(a). These representative step changes are the

very fact of what was called “step change” in this study and these were entered to the

simulator.

Feed rate 15% decrease - open loop operation

Experimental details: In the first experiment the standpipe slide valve was set to manual

mode, equal to its average opening of the previous one hour steady state operation. The time

this action was taken is marked with the first vertical dotted line in diagrams of Fig. 5.

Average values in non-linear systems produce momentary instabilities; hence a line-out

period of 10 min was needed. When the unit reached steadiness (second vertical dotted line in

diagrams of Fig. 5), a step change in the feed mass flow was imposed (from 15.5 to 12.8

gr/min), which produced the transient in feed flow rate shown in Fig. 5(a). The feed flow rate

that was entered to the simulator represented this transient by the consecutive step changes

shown in Fig. 5(a). After a period of 20 min, the pattern of changes of Fig. 5(a) was manually

imposed to the stripper slide valve. The specific oscillatory pattern was chosen to distinct the

effect of catalyst circulation rate, without considerably alter the pressure balance of the unit.

Effect of change in feed rate: The decrease in feed rate caused an instantaneous increase in

the riser temperature (Fig. 5(b)), because less feed consumed less heat for vaporization.

Furthermore, the catalyst to oil ratio increased, since the catalyst circulation rate was the same

18

(open loop operation) and the feed rate lower. The latter resulted in higher conversion and

coke yield (feed basis) observed and predicted, as shown in Table 4. However, the coke rate

entering the regenerator decreased, because the feed rate was lower. This produced a decrease

in the regenerator temperature (Fig. 5(b)). As shown Fig. 5(c) the lower coke rate entering the

regenerator resulted in a decrease in the CO2 and SO2 concentration and a parallel increase of

the excess O2.

Effect of change in catalyst flow: The comparison of Fig. 5(a), (b) and (c) shows that the

oscillation in the catalyst flow rate, produced by the changes in the stripper slide valve

opening, showed a more pronounced effect on flue gas composition than on regenerator

temperature. Evidently, the many factors involved in the regenerator heat balance

(temperature and flow of catalyst and gas entering and exiting the regenerator, dense bed

volume, heat loss) result in smoother transients of the regenerator temperature. The decrease

in regenerator temperature led to a smaller decrease in standpipe temperature (Fig. 5(e)),

which again resulted to an even smaller decrease in riser temperature (Fig. 5(b)). The loop

was continued for a period of 100 min (third vertical dotted line in diagrams of Fig. 5), until

convergence. The variation of the stripper slide valve opening led to implicit variation of the

stripper pressure drop (Fig. 5(d), without a parallel variation of the regenerator bed height.

The reason for this is the much larger regenerator diameter. The change in regenerator

temperature and the small variation of the catalyst flow rate resulted in the temperature

profiles of liftline and standpipe shown in Fig. 5(e). The simulator predictions are in good

agreement with the real dynamic behavior of the unit. The form of entering the step change in

five representational consecutive steps has a negligible effect in the simulation of the unit

dynamic responses. In open loop operation the simulator can predict the dynamic behavior of

the pilot plant in terms of temperature, yield of combustion reactions, and pressure, which are

the variables that can be measured on-line.

19

Feed rate 15 % decrease - closed loop operation

Experimental details: In this case the previous experiment was repeated (Fig. 6(a)), but for

closed loop operation of the unit. The standpipe slide valve was set to control the riser

temperature and the stripper slide valve operated for constant stripper pressure drop.

Accordingly, the routine of the simulator that adjusts the catalyst circulation for constant riser

temperature was activated, while the catalyst rates at the entrance and the exit of the stripper

were set equal. The simulation results are of course quite different from the actions of the

controllers, which are much slower. Thus, the results of the simulator are examined in the

sense of an “ideal efficiency controller”. Nonetheless, the steady states and the general trends

of the simulator and the pilot plant should be similar.

Effect of change in feed rate: The decrease in feed rate should have produced an increase in

riser temperature, though the controller (or the adjustment routine) lowered the catalyst

circulation rate to satisfy the heat balance of the riser for temperature equal to 526.8°C. The

first regenerator response predicted by the simulator is a minor temperature increase, owed to

the lower cold catalyst mass entering- and hot catalyst mass exiting the regenerator.

Thereafter, the regenerator temperature decreased due to the lower coke rate that decreased

the exothermic combustion reactions. As shown in Fig. 6(b), the phenomenon was quite

different in the unit. The standpipe slide valve controller is not efficient enough to balance the

catalyst circulation rapidly, thus it produced an oscillation in the riser temperature for a period

of 90 min. Moreover, the stripper bed height oscillated around its average value (Fig. 6(c)),

whereas the regenerator bed height was again relatively constant, owing to its larger diameter.

As shown in Fig. 6(c) and (d) the results of the simulator are straightforwardly comparable

with the results of the pilot plant, yet the simulator is “faster”. The final steady state was

achieved after 100 min in the unit, while a transient of only 40 min is predicted by the

simulator. This is a typical example of how such a simulator can help towards the direction of

20

optimal unit control, if for instance the demand was for constant riser temperature and

minimum transition time.

Feed preheat temperature 130% increase - open loop operation

Experimental details: In this case the effect of increasing the feed preheat temperature from

104°C to 232°C on the dynamic performance of the pilot plant in open loop operation was

explored. At the time marked with the first vertical dotted line in diagrams of Fig. 7, the

standpipe slide valve was set to manual mode. After a line-out period of 30 min (second

vertical dotted line in diagrams of Fig. 7), the step change was imposed to feed preheat zones

1 and 2, which produced the transient in feed preheat temperature shown in Fig. 7(a). The

long transient of preheat change was represented, for the simulator needs, by the consecutive

step changes shown in Fig. 7(a). In this case the stripper slide valve opening was set constant.

Effect of change in feed preheat: The increase in feed preheat temperature caused an instant

increase in the riser temperature as the heat balance of the riser imposes (Fig. 7(b)). This

increase in riser temperature led to higher feed conversion but not different coke yield (Table

5), since coke production is not significantly influenced by temperature. The method of

representing the transient of change in feed preheat with consecutive steps is less accurate in

this case. However, the general trends of the pilot plant and the simulator are similar (Fig.

7(b)). As shown in Fig. 7(b) and (e), there is a deviation of 2-5°C in the prediction of

regenerator and standpipe temperature. This is not followed by a difference in the flue gas

composition, which was predicted and observed relatively constant (Fig. 7(c)). Effects of wall

heaters that operate for the establishment of pseudo-adiabatic conditions could have generated

this increase in regenerator and standpipe temperature and could not be depicted by the

simulator. The pressure drop of the stripper and the regenerator were measured and predicted

constant (Fig. 7(d)), as the catalyst circulation rate and coke yield did not significantly

21

change. As presented in Table 5, the accuracy in the prediction of the final steady state in this

case is fair.

Feed preheat temperature 130% increase - closed loop operation

Experimental details: The previous experiment was repeated, but in closed loop operation

of the pilot plant (Fig. 8(a)). The slide valves controllers were set to automatic operation and

the routine of the simulator that adjusts the catalyst circulation for constant riser temperature

was activated. The imposed change produced the transient of Fig. 8(a), which again was

represented by five consecutive step changes.

Effect of change in feed preheat: The inefficiency of the standpipe slide valve controller

produced an oscillation in riser temperature (Fig. 8(b)), whereas the sluggish behavior of the

stripper slide valve produced large oscillation in the stripper bed height and even a small

oscillation in the regenerator bed height (Fig. 8(d). The fluctuation of the catalyst stream

entering the regenerator resulted in hard oscillation in the flue gas composition (Fig. 8(c)).

The final steady state was achieved at a lower catalyst circulation rate and coke yield, with

higher regenerator temperature as shown in Table 5. The experimental results presented in

Table 5 are of moderate accuracy, as in this case the final steady state was not well

established. The performance of the simulator is again “faster”, yet in this case the difference

between experiment and model is much more crucial. More robust control could definitely

enhance the steadiness in the operation of the unit and would produce smother profiles in

critical unit variables as the riser temperature and the regenerator flue gas. The response of

the simulator proposes that it could be possible with optimal control to set the regenerator to

absorb the increase in the feed preheat temperature and preserve the riser temperature

constant at 526°C with smooth unit performance.

22

Conclusions

A dynamic simulator of the FCC integrated riser-regenerator system was presented. The

nonlinear dynamic and multivariable model was verified with a set of dynamic experiments

performed in the pilot plant of CPERI. The term "dynamic experiments" is used to express

experiments, in which a step change is imposed to a manipulated process variable, while

recording the transient of a number of process variables from the current steady state to the

new steady state the system will eventually reach, in both open and closed loop operation.

The simulator performs satisfactorily, in describing the complex responses of the unit to

typical disturbances. Increases in the riser input variables corresponded to aggressive

responses of the system, since the riser has a very small contribution to the dynamic behavior

of the integrated system. However, after the immediate new state was reached, the regenerator

led the system to the new steady state in much greater times. The excellent convergence

between observed and predicted values for reactor and regenerator temperatures indicates the

accurate formulation of the mass and energy balances. The results of both the simulator and

the pilot plant are in excellent agreement with the experience of real-time operation of FCC

units. The accurate simulation of the pilot plant is significant for process optimization studies.

The ultimate scope of this study is to utilize the simulator as a basis for model based control

of the pilot plant. The purpose of such a controller would be to provide the operator with the

ability of driving the pilot process to desired states, such as the maximization of the

selectivity of a desired product or the maximization of the process conversion.

Acknowledgement

Financial support by the Ministry of National Education and Religious Affairs (program

2.2.01 ARCHIMIDES I, EPEAEK II) and the General Secretariat of Research and

Technology Hellas (program AKMON 01) is gratefully acknowledged.

23

Nomenclature

A cross-sectional area (m2)

c, cx total coke yield wt%, catalytic coke yield wt%

CA aromatic carbon content wt% of feed

CCR conradson carbon residue wt% of feed

D diameter (m)

dp catalyst particle mean diameter (m3)

cp specific heat (kcal/molK)

Ex, Ec activation energy of reaction to x, c (kcal mol-1)

fb, fe bubble, emulsion phase volume fraction

Fib molar rate in bubble (mol s-1)

cie molar concentration in emulsion (mol s-1)

Fif molar rate in freeboard (mol s-1)

Ht bubble-emulsion heat interchange (kcal m-3 s)

KH heat interchange rate group (kcal m-3 s)

KMi mass interchange rate group (mol m-3 s)

KRjb reaction rate group of reaction j - bubble phase (mol m-3 s)

KRje reaction rate group of reaction j - emulsion phase (mol m-3 s)

KRjf reaction rate group of reaction j - freeboard (mol m-3 s)

KSV1, KSV2 characteristic constant of slide valve SV1, SV2

Kti bubble-emulsion gas interchange coefficient (s-1)

kx, kc pre-exponential factor of reaction to x, c

L height (m)

l dimentioneless height

MWF feed molecular weight

NC carbon number in the average feed molecule

24

nx, nc catalyst decay exponent of reaction to x, c

NN nitrogen number in the average feed molecule

NT total nitrogen content wt% of feed

Qb enthalpy rate in bubble phase (kcal s-1)

QC enthalpy rate of catalyst (kcal s-1)

Qge enthalpy rate of gas in emulsion phase (kcal s-1)

Qloss heat loss from the dense bed (kcal s-1)

P pressure (Pa)

S sulfur content wt% of feed

Te temperature of emulsion phase (kcal s-1)

TRX riser reactor temperature (°F)

tC catalyst residence time (s)

tdead time lug in standpipe or liftline (s)

ug superficial gas velocity (m s-1)

ut catalyst particle terminal velocity (m s-1)

V volume (m3)

WHSV weight hourly space velocity (hr-1)

CW catalyst circulation rate (kg s-1)

FW gas-oil feed rate (kg s-1)

x gas-oil conversion wt%

ΔHcrack heat of catalytic cracking (kcal kg-1)

ΔHRj heat of reaction j (kcal mol-1 )

Greek Letters

αij stoichiometric coefficient of component i in rection j

εb voidage of bulk catalyst

εmf voidage at minimum fluidization

25

εe dense bed emulsion void fraction

εf freeboard void fraction

εr riser void fraction

ρp catalyst density (kg m-3)

Subscripts

g gas phase

b bubble phase

e emulasion phase

f dilute phase

Unit Section Subscripts

D dense phase or bottom section

C cone intermediate section

F dilute phase or top section

CY regenerator cyclone

RS riser

RG regenerator

ST stripper

SP standpipe

LL liftline

26

Appendix A: Hydrodynamic Correlations and Pressure Balance

Simulation of Riser

The weight hourly space velocity (WHSV) and the solids residence time (ts) were calculated

by eqs.(32) and (33), following the pilot riser geometry, that is divided in three regions:

( ) ( ) (( ))

:RS

:RS :RS :RS :RS :RS :RS1 1 1F

p D D C C F F

WWHSVV V Vρ ε ε ε

=− + − + −

(32)

:RS:RS

:RS

3600

FC

C

WtWHSV W

= (33)

(a) The mixing region at the riser bottom. The void fraction ( :RSDε ) and subsequently the

catalyst inventory of this region, were related to the superficial gas velocity by means of the

empirical correlation of Richardson and Zaki (Richardson & Zaki, 1954) (eq.(34)), which

substantiates for a dense regime in the bottom region of the pilot unit.

1

:RS:RS

:RS

/ zg:D

Dt

uu

ε⎛ ⎞

= ⎜ ⎟⎝ ⎠

(34)

(b) The conical shaped intermediate region. Because of the very small volume of the

intermediate region (15% of total riser volume), a simple approximation of averaged (between

top and bottom regions) hydrodynamic attributes was used (Pugsley & Berruti, 1996).

(c) The fast fluidization region at the riser top, which was simulated under the following

assumptions: (i) the flow is fully developed, thus its hydrodynamic features remain constant

with height; (ii) the total volumetric yield of the reaction is flowing through the whole height

of this region; (iii) the particle acceleration is considered to be negligible. Hence, eq.(35)

holds:

: :RS :RS:RS

:RS :RS : :RS :RS

g F p FF

F C g D p F

u Ay W u A

ερ

ρ=

+ (35)

In eq.(35) :RSFy is the average gas-solids slip factor for the top section of the riser, which

was proven to play an important role in small diameter riser reactors (Bollas et al., 2002). The

27

correlation of Pugsley et al. (Pugsley & Berruti, 1996) was applied for the estimation of the

gas-solids slip factor as shown in eq.(36), where Frg and Frt are the Froude numbers for the

superficial gas velocity and solids terminal velocity, respectively:

0.41:RS : :RS2

: :RS

5.61 0.47Fg F

yFr

= + + t FFr (36)

A detailed pressure gradient analysis is required for small diameter risers (Bollas et al.,

2002). For this analysis, all pressure gradients should be taken into account, and eq.(37) is

valid where ΔPfg is the gas-wall frictional pressure drop, ΔPfs is the solids-wall frictional

pressure drop, ΔPacc is the pressure drop due to solids acceleration, and the other terms

represent the pressure drop due to solids and gas static head throughout the total riser height:

( )RS :RS :RS :RS RS :RS RS RS RS1fg fs acc g pP P P P gL gLε ρ ε ρΔ = Δ + Δ + Δ + + − (37)

Simulation of Regenerator

For group A particles the emulsion gas superficial velocity is the gas velocity for zero net

flow of solids, which equals the minimum bubbling velocity, plus (concurrent gas/solids

flows) or minus (countercurrent gas/solids flow) the superficial solids velocity in the

emulsion phase:

ge mbu u use= ± (38)

For the evaluation of the minimum fluidization velocity the equation of Wen and Yu (Kunii

& Levenspiel, 1977) is applied. For group A particles the minimum bubbling velocity, umb, is

evaluated by the correlation of Abrahamsen and Geldart (Abrahamsen & Geldart, 1980),

which considers the effect of catalyst fines, , on uf mb:

( )

( )0.126 0.523

0.9340.8 0.934

2300 exp 0.716ge gemb

mf p p ge

uu d g

ρ μ

ρ ρ=

−

f (39)

The superficial gas velocity in the dense zone is then obtained by eq.(40):

:RGg gb geu u u= + (40)

28

The fraction of the bubbles in the dense zone is:

gbb

b

uf

υ= (41)

The absolute bubble rise velocity bυ is calculated as a function the isolated bubble rise

velocity:

( )0.5:RG0.711b b ggd u uυ ge= + − (42)

The bubble diameter is estimated by the Wen-Mori correlation (Kunii & Levenspiel, 1977):

( )

( ) ( )

1:RG

1 0:RG

0.3expD

D D

lb b D

l lDb b

d d L lDd d

=

= =

⎛ ⎞−= −⎜

− ⎝ ⎠D ⎟ (43)

The initial bubble diameter and the maximum bubble diameter are estimated by eqs.(26)

and (27), respectively:

( ) ( )(0.4

0 0:RG0.2

1.38 11000

D Dl lb gd u

g= =⎛= ⎜

⎝ ⎠)mbu ⎞− ⎟ (44)

( ) ( )( )( )( )( )22.7

0.41 1

:RG :RGmin 0.652 , 2p

D D

dtl l

b D g mb

ud A u u

g= =

⎡ ⎤⎢ ⎥

= −⎢ ⎥⎢ ⎥⎣ ⎦

(45)

The emulsion to freeboard elutriation rate of a fraction of particles with average

diameter d

*iK

pi is evaluated by the Zenz and Weil correlation (Geldart, 1985). The total

entrainment rate is then obtained by adding the rates of each respective fraction of

particles. The catalyst density in the freeboard is a function of the gas-solids slip velocity,

which is calculated on the basis of the correlation of Patience et al. (Patience et al., 1992), as

shown in eq.

*tK

(46):

0.411 5.6 / 0.47gf

sfgf t

uu

Fr Fr=

+ + (46)

The freeboard voidage is then calculated by eq.(47):

29

*

1 tf

p sf

Ku

ερ

= − (47)

The pressure drop throughout the regenerator is calculated from the solids static head as

shown in eq.(48):

( ) ( )RG :RG :RG1 1p e e D p f FP f gL gLρ ε ρ εΔ = − + − (48)

Simulation of Stripper, Standpipe, Liftline and Slide Valves

The pressure drop throughout the stripper is calculated from the solids static head as shown

in eq.(49):

( )ST :ST1p mf DP ρ εΔ = − gL (49)

The catalyst circulation throughout the unit is regulated by two slide valves, one at the riser

entrance and one at the stripper exit. The catalyst circulation rate at the entrance and exit of

the regenerator was correlated with the slide valves opening and pressure drop by eq.(50)

(Judd & Dixon, 1978):

( )(0.52 2 0.50 :SV

SV 2 20 :SV

2 1tC p

t

A AW k PA A

ρ ε⎛ ⎞

= ⎜ ⎟−⎝ ⎠)SV− Δ (50)

Appendix B: The bulk molecular feedstock characterization approach

The effect of feedstock quality in eqs.(1) and (2), was simulated with a “bulk molecular

characterization approach” (Bollas et al., 2004), that is the calculation of the relative

(compared to a reference feedstock) potential of an FCC feedstock to enhance catalytic

cracking conversion (crackability) and coke formation (coking tendency). The proposed

models focus on the needs of industry; hence they include the bulk feedstock properties of

Fig. 9, measured with standard analytical procedures accessible to the average

refinery (Bollas et al., 2004). Fig. 9 presents the computational methodology for the

breakdown of an FCC feedstock into pseudo-components and for the estimation of the

30

properties of light and heavy fractions, separately. This splitting and lumping scheme was

applied to explore the different extent of contributions to the catalytic cracking of the heavy

and the light fractions.

The bulk feedstock properties were combined properly to deliver five functional groups to

characterize the behavior of FCC feedstocks at cracking conditions, which involved: the

prediction of catalyst poisoning, the estimation of the cracking extent, the estimation of

coke/conversion selectivity, and two coking precursors for contaminant and residue coke

formation (Bollas et al., 2004), as shown in eqs.(51) and (52):

( ) ( ) :RS1 2 3 exp

100

xnC

C C A NRX

tx w N N C w N wx W

⎛ ⎞−= − + +⎡ ⎤ ⎜⎣ ⎦− ⎝ ⎠

xEHSV RT ⎟ (51)

( ) ( ) ( )2 :RS

4 5 1 2 3

exp0.437

expc

CCR N N

nC c

A C C A NRX

NTc k CCR k NNT S

t Ew w C w N N C w N wWHSV RT

−⎛ ⎞= + ⎜ ⎟+⎝ ⎠⎛ ⎞−⎡ ⎤ ⎡ ⎤+ + − + + ⎜ ⎟⎣ ⎦ ⎣ ⎦ ⎝ ⎠

(52)

A large database of experiments performed in the FCC pilot plant of CPERI, with a

standard catalyst and a large variety of feedstock qualities, was used for the development and

validation of eqs.(51) and (52). Moreover, a database of experiments with different catalysts

and feedstocks was used for the assignment of a “catalyst index” to each catalyst to describe

the effect of catalyst quality in eqs.(1) and (2). The description of catalyst activity and

selectivity with a simple index is a common strategy in industry, in order to validate catalyst

performance in commercial units. The values of the parameters of eqs.(51) and (52) are given

in Table 1.

31

Literature Cited

Abrahamsen, A.R., Geldart, D., (1980). Behaviour of gas-fluidized beds of fine powders, part

I. Homogeneous expansion. Powder Technology, 26(1 , May-Jun. 1980), 35-46.

Ali, H., Rohani, S., (1997). Dynamic modeling and simulation of a riser-type Fluid Catalytic

Cracking Unit. Chemical Engineering & Technology, 20(2), 118-130.

Ali, H., Rohani, S., Corriou, J.P., (1997). Modelling and control of a riser-type Fluid Catalytic

Cracking (FCC) Unit. Chemical Engineering Research & Design, 75(A4), 401-412.

Arbel, A. et al., (1995). Dynamic and control of Fluidized Catalytic Crackers. 1. Modeling of

the current generation of FCCs. Industrial & Engineering Chemistry Research, 34(4), 1228-

1243.

Arbel, A., Rinard, I.H., Shinnar, R., (1995). Dynamics and control of Fluidized Catalytic

Crackers. 2. Multiple steady-states and instabilities. Industrial & Engineering Chemistry

Research, 34(9), 3014-3026.

Arthur, J.R., (1951). Reactions between carbon and oxygen. Transactions of the Faraday

Society, 47, 164-178.

Blanding, F.H., (1953). Reaction rates in the catalytic cracking of petroleum. Industrial &

Engineering Chemistry, 45, 1186.

Bollas, G.M. et al., (2002). Modeling small-diameter FCC riser reactors. A hydrodynamic and

kinetic approach. Industrial & Engineering Chemistry Research, 41(22), 5410-5419.

Bollas, G.M. et al., (2004). Bulk Molecular Characterization Approach for the simulation of

FCC feedstocks. Industrial & Engineering Chemistry Research, 43(13), 3270-3281.

Cristea, M.V., Agachi, S.P., Marinoiu, V., (2003). Simulation and model predictive control of

a UOP fluid catalytic cracking unit. Chemical Engineering And Processing, 42(2), 67-91.

Davidson, J.F., Clift, R., Harrison, D. (1985). Fluidization. Academic Press Inc., London.

Elnashaie, S., Elshishini, S.S., (1993). Digital-Simulation of industrial Fluid Catalytic

Cracking Units. 4. Dynamic behavior. Chemical Engineering Science, 48(3), 567-583.

32

Elnashaie, S., Abasaeed, A.E., Elshishini, S.S., (1995). Digital-Simulation of industrial Fluid

Catalytic Cracking Units. 5. Static and dynamic bifurcation. Chemical Engineering Science,

50(10), 1635-1644.

Faltsi-Saravelou, O., Vasalos, I.A., (1991). Fbsim - A model for fluidized-bed simulation. 1.

Dynamic modeling of an adiabatic reacting system of small gas-fluidized particles.

Computers & Chemical Engineering, 15(9), 639-646.

Faltsi-Saravelou, O., Vasalos, I.A., Dimogiorgas, G., (1991). Fbsim - A model for fluidized-

bed simulation. 2. Simulation of an industrial fluidized catalytic cracking regenerator.

Computers & Chemical Engineering, 15(9), 647-656.

Geldart, D., (1973). Types of gas fluidization. Powder Technology, 7(5), 285-292.

Geldart, D. (1985). Elutriation, in Davidson, J.F., Clift, R. & Harrison, D. (Eds.),

Fluidization. Academic Press Inc., London, p. 383.

Han, I.S., Chung, C.B., (2001a). Dynamic modeling and simulation of a fluidized catalytic

cracking process. Part I: Process modeling. Chemical Engineering Science, 56(5), 1951-1971.

Han, I.S., Chung, C.B., (2001b). Dynamic modeling and simulation of a fluidized catalytic

cracking process. Part II: Property estimation and simulation. Chemical Engineering Science,

56(5), 1973-1990.

Hernandez-Barajas, J.R., Vazquez-Roman, R., Salazar-Sotelo, D., (2006). Multiplicity of

steady states in FCC units: effect of operating conditions. Fuel, 85(5-6), 849-859.

Howard, J.B., Williams, G.C., Fine, D.H. (1973). Kinetics of Carbon Monoxide Oxidation in

Post Flame Gases. In 14th Symposium (International) on Combustion, Pittsburgh, The

Combustion Institute.

Jacob, S.M. et al., (1976). Lumping and reaction scheme for Catalytic Cracking. Aiche

Journal, 22(4), 701-713.

Judd, M.R., Dixon, P.D., (1978). Flow of fine, dense solids down a vertical standpipe. AIChE

Symp Ser, 74(176), 38-44.

33

Kesler, M.G., Lee, B.I., (1976). Improve prediction of enthalpy of fractions. Hydrocarbon

Processing, 55(3), 153-158.

Kunii, D., Levenspiel, O. (1977). Fluidization Engineering. Robert E. Krieger Publishing

Company Inc., Florida.

Lee, E., Groves, F.R., Jr., (1985). Mathematical model of the Fluidized Bed Catalytic

Cracking Plant Trans. Soc. Comput. Simul. Int., 2(3), 219-236

Lopez-Isunza, F., (1992). Dynamic modeling of an industrial Fluid Catalytic Cracking Unit.

Computers & Chemical Engineering, 16, S139-S148.

McFarlane, R.C. et al., (1993). Dynamic simulator for a Model-IV Fluid Catalytic Cracking

Unit. Computers & Chemical Engineering, 17(3), 275-300.

Morley, K., De Lasa, H.I., (1987). On the determination of kinetic parameters for the

regeneration of cracking catalyst. Canadian Journal of Chemical Engineering, 65(5), 773-

777.

Patience, G.S. et al., (1992). Scaling considerations for circulating fluidized bed risers.

Powder Technology, 72(1), 31-37.

Pugsley, T.S., Berruti, F., (1996). A predictive hydrodynamic model for circulating fluidized

bed risers. Powder Technology, 89(1), 57-69.

Richardson, J.F., Zaki, W.N., (1954). Sedimentation and Fluidization. I. Transactions of the

Institution of Chemical Engineers, 32, 35.

Tone, S., Miura, S.I., Otake, T., (1972). Kinetics of oxidation of coke on silica-alumina

catalysts. Bull. Jap. Petrol. Inst., 14(1), 76.

Voorhies, A., (1945). Carbon formation on catalytic cracking. Industrial & Engineering

Chemistry, 37, 4.

Wang, G.-x. et al., (1986). Kinetics of combustion of Carbon and Hydrogen in carbonaceous

deposits on zeolite-type cracking catalysts. Industrial & Engineering Chemistry, Process

Design and Development, 25(3), 626-630.

34

35

List of Tables

Table 1: Parameters of the model of the pilot riser

Table 2: Parameters of the model of the regenerator

Table 3: Bulk properties of the feedstock and the catalyst used in the experiments examined

Table 4: Steady state results (experimental and predicted) of open and closed loop behavior of

the pilot plant for a 15% decrease in feed rate

Table 5: Steady state results (experimental and predicted) of open and closed loop behavior of

the pilot plant for a 130% increase in feed preheat temperature

36

Table 1: Parameters of the model of the pilot riser

Operating Conditions Effect of Feedstock Heat of Cracking

kx 200.04 w1 0.0261 α1 3.49

Ex 8.9 w2 5.6899 α2 18.6

nx -0,78 w3 0.6512 α3 0.021

kc 1.283 w4 0.0317 b1 -9.50

Ec 0.9 w5 0.0442 b2 53.4

nc -0.90 kCCR 0.3000 b3 0.044

kN 0.3214

37

Table 2: Parameters of the model of the regenerator

Frequency Factor Activation Energy (kcal mol-1)

Reference

k01+k02 1.4E05 Ε1 29.9 (Morley & De Lasa, 1987)

k01/k02 2.5E03 Ε2 12.4 (Arthur, 1951)

k03 1.3E03 Ε3 30.0 (Howard, Williams & Fine, 1973)

k04 3.5E03 Ε4 13.8 (Tone, Miura & Otake, 1972)

k05 3.3E07 Ε5 37.7 (Wang et al., 1986)

k06 1.4E05 Ε6 29.9 (Faltsi-Saravelou, Vasalos & Dimogiorgas, 1991)

38

Table 3: Bulk properties of the feedstock and the catalyst used in the experiments examined

Feedstock Properties Catalyst Properties

Code Name #19 Code Name #43

Gravity (API) 18.9 Bulk Density (kg m-3) 900

Refractive index (at 20°C) 1.5226 Mean Particle Diameter (μ) 75

Sulfur %wt 2.58 Al2O3, wt% 39.1

Nitrogen %wt 0.13 SiO2, wt% 59.6

Carbon %wt 85.3 Re2O3, wt% 0.65

Con. carbon residue %wt 0.36 Fe, wt% 0.59

TBP distillation (°C) Particle size distribution

IBP 303.6 fraction (wt%) size (μ)

10% 379.2 0

30% 422.9 10

50% 454.4 50

70% 483.1 80

90% 524.2 90

FBP 551.6 100

39

Table 4: Steady state results (experimental and predicted) of open and closed loop behavior of

the pilot plant for a 15% decrease in feed rate

open loop behavior closed loop behavior

case examined steady state 1 steady state 2 steady state 1 steady state 2

feed rate (kg s-1) 25.19E-3 20.67E-3 25.64E-3 20.88E-3

feed preheat (°C) 104.4 104.5 104.4 104.4

experimental vs. predicted operational variables

experimental vs. predicted operational variables

catalyst to oil ratio 15.6 15.6 18.9 18.9 13.6 13.7 17.48 14.7

riser temperature (°C) 526.8 535.5 534.7 526.8 526.8

reg. temperature (°C) 683.3 683.7 679.4 679.3 683.3 683.6 681.7 680.4

experimental vs. predicted yields experimental vs. predicted yields

conversion wt% on feed 66.8 65.8 69.6 70.7 64.7 63.7 68.4 65.8

coke yield wt% on feed 5.78 5.77 6.63 6.66 5.31 5.32 6.49 5.65

carbon wt% on reg. cat.

0.035 0.030 0.020 0.21 0.013 0.019 0.015 0.012

40

Table 5: Steady state results (experimental and predicted) of open and closed loop behavior of

the pilot plant for a 130% increase in feed

open loop behavior closed loop behavior

case examined steady state 1 steady state 2 steady state 1 steady state 2

feed rate (kg s-1) 25.27E-3 25.25E-3 25.03E-3 25.26E-3

feed preheat (°C) 104.4 232.4 104.4 232.3

experimental vs. predicted operational variables

experimental vs. predicted operational variables

catalyst to oil ratio 14.8 14.7 15.2 14.5 13.5 13.5 12.9 11.21

riser temperature (°C) 526.8 540.6 540.9 526.7 526.7

reg. temperature (°C) 688.1 688.8 696.8 693.9 685.4 685.3 694.4 693.3

experimental vs. predicted yields experimental vs. predicted yields

conversion wt% on feed 64.7 64.7 64.9 66.4 65.2 63.53 60.5 60.28

coke yield wt% on feed 5.37 5.56 5.57 5.52 5.07 5.28 5.11 4.72

carbon wt% on reg. cat.

0.030 0.037 0.025 0.033 0.010 0.023 0.015 0.013

41

List of Figures

Fig. 1: Schematic diagram of the FCC pilot plant of CPERI.

Fig. 2: Physical model of the two-phase regime used for the simulation of the regenerator.

Fig. 3: Logical scheme of the dynamic simulator of the FCC pilot plant of CPERI.

Fig. 4: Structure of solution sequence of the integrated FCC simulator.

Fig. 5: Open loop responses of pilot plant and simulator for a 15% decrease in feed rate.

Fig. 6: Closed loop responses of pilot plant and simulator for a 15% decrease in feed rate.

Fig. 7: Open loop responses of pilot plant and simulator for a 130% increase in feed preheat

temperature.

Fig. 8: Closed loop responses of pilot plant and simulator for a 130% increase in feed

preheat temperature.

Fig. 9: Logical scheme of the proposed method for the simulation of the effect of feed

properties and operating conditions on the coke yield of the FCC process.

42

Fig. 1: Schematic diagram of the FCC pilot plant of CPERI.

43

Fig. 2: Physical model of the two-phase regime used for the simulation of the regenerator.

44

Fig. 3: Logical scheme of the dynamic simulator of the FCC pilot plant of CPERI.

45

Fig. 4: Structure of solution sequence of the integrated FCC simulator.

46

Fig. 5: Open loop responses of pilot plant and simulator for a 15% decrease in feed rate.

47

Fig. 6: Closed loop responses of pilot plant and simulator for a 15% decrease in feed rate.

48

Fig. 7: Open loop responses of pilot plant and simulator for a 130% increase in feed preheat

temperature.

49

Fig. 8: Closed loop responses of pilot plant and simulator for a 130% increase in feed preheat

temperature.

50

Fig. 9: Logical scheme of the proposed method for the simulation of the effect of feed

properties and operating conditions on the coke yield of the FCC process.

51